1 / 22

Prof . Congduc Pham http://www.univ-pau.fr/ ~cpham Université de Pau, France

Network lifetime and stealth time of wireless video sensor intrusion detection systems under risk-based scheduling. Prof . Congduc Pham http://www.univ-pau.fr/ ~cpham Université de Pau, France. ISWPC, 2011 Hong-Kong Wednesday, February 23 rd. Wireless Video Sensors. Imote2.

seven
Download Presentation

Prof . Congduc Pham http://www.univ-pau.fr/ ~cpham Université de Pau, France

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Network lifetime and stealth time of wirelessvideosensor intrusion detectionsystemsunderrisk-basedscheduling Prof. Congduc Pham http://www.univ-pau.fr/~cpham Université de Pau, France ISWPC, 2011 Hong-Kong Wednesday, February 23rd

  2. Wireless VideoSensors Imote2 Multimedia board

  3. Surveillance scenario (1) • Randomlydeployedvideosensors • Not onlybarriercoverage but general intrusion detection • Most of the time, network in so-calledhibernate mode • Most of active sensornodes in idle modewithlow capture speed • Sentrynodeswithhigher capture speed to quicklydetect intrusions

  4. Surveillance scenario (2) • Nodesdetecting intrusion must alert the rest of the network • 1-hop to k-hopalert • Network in so-calledalerted mode • Capture speed must beincreased • Ressources shouldbefocused on makingtracking of intruderseasier

  5. Surveillance scenario (3) • Network should go back to hibernate mode • Nodes on the intrusion path must keep a high capture speed • Sentrynodeswithhigher capture speed to quicklydetect intrusions

  6. Wholeunderstanding of the sceneiswrong!!! Don’t miss important events! Real scene Whatiscaptured

  7. How to meet surveillance app’s criticality • Capture speed canbe a « quality » parameter • Capture speed for node v shoulddepend on the app’scriticality and on the level of redundancy for node v • Note thatcapturing an image does not meantransmittingit • V’scapture speed canincreasewhen as V has more nodescoveringitsownFoV - coverset

  8. RedundancyNode’scover set • Each node v has a Field of View, FoVv • Coi(v) = set of nodes v’ such as v’Coi(v)FoVv’ covers FoVv • Co(v)= set of Coi(v) V2 V1 V4 V V3 Co(v)={V1,V2,V3,V4}

  9. Criticality model (1) • Link the capture rate to the size of the coverset • High criticality • Convexshape • Most projections of x are close to the max capture speed • Lowcriticality • Concave shape • Most projections of x are close to the min capture speed • Concave and convexshapesautomaticallydefinesentrynodes in the network

  10. Criticality model (2) • r0canvary in [0,1] • BehaViorfunctions (BV) defines the capture speed according to r0 • r0 < 0.5 • Concave shape BV • r0 > 0.5 • Convexshape BV • We propose to use Beziercurves to model BV functions

  11. Some typical capture speed • Set maximum capture speed: 6fps or 12fps for instance • Nodeswithcoverset size greaterthan N capture at the maximum speed N=6 P2(6,6) N=12 P2(12,3)

  12. How to build an intrusion detection system • Static • Prior to deployment, define r° in [0,1] according to the application’scriticality • Risk-based • R0is set initiallylow : R°min • Somenodes serve as sentrynodes • On intrusion, increase R° to R°maxduring an givenalertperiod (Ta) • After Ta, go back to R°min • 2 variants • R°moves fromR°minto R°max in one step • R°moves fromR°minto R°maxby reinforcementbehavior

  13. Risk-basedscheduling in images (1) • R°=R°min=0.1, R°max=0.9, no alert |Co(vj)|=3 fps=0.14 0.81 |Co(vi)|=8 0.14 fps=0.81 3 8

  14. Risk-basedscheduling in images (2) • R°R°=R°max=0.9 |Co(vj)|=3 Fps=1.9 2.8 1.9 |Co(vi)|=8 Fps=2.8 3 8

  15. Simulation settings • OMNET++ simulation model • Videonodes have communication range of 30m and depth of view of 25m, AoVis 36°. 150 sensors in an 75m.75m area. • Battery has 100 units, 1 image = 1 unit of batteryconsumed. • Max capture rate is 3fps. 12 levels of cover set. • Full coverageisdefined as the regioninitiallycoveredwhen all nodes are active

  16. meanstealthtime (MST) t1-t0is the intruder’sstealth time velocityis set to 5m/s t0 t1 • intrusions startsat t=10s • when an intruderisseen, computes the stealth time, and starts a new intrusion until end of simulation

  17. meanstealthtimestaticscheduling

  18. meanstealth timerisk-basedscheduling 450s 1300s • Sensornodesstartat 0.1 thenincrease to 0.9 if alerted (by intruders or neighbors) and stayalerted for Ta seconds

  19. meanstealth timew/woreinforcement Ir=0.6 • On alert 0.1Ir, then • 2 alertmsg IrIr+1 • Until Ir=R°max • Reinforcementalwaysincreases the network lifetime • Meanstealth time is close to the no-reinforcement case, especiallywhen Ta>20s

  20. Withreinforcementvarious initial threshold • Ir=0.4 • Ir=0.5 • Ir=0.6 • Reduce Iralwaysincreases the network lifetime • For small value of Ta, MST increaseisnoticeable • It isbetter to increase Irthanincrease Ta.

  21. 0 0 <5 <5 <10 <10 <15 <15 >15 >15 Sentry nodes # of cover sets # intrusion detected

  22. Conclusions • Models the application’scriticality as beziercurves and schedules the videonode capture rate according to the redundancylevel • Withthis model, a risk-basedschedulingcanincrease the network lifetimewhilemaintaining a highlevel of service (meanstealth time) • Reinforcementbehaviorisbeneficial and itisbetter to keep the alertperiodlow <=20s for instance

More Related